Recovering the History of Filtered JPEG Images for Forensic Detection

Abstract

The image editing software available today are apt for creating visually compelling and sophisticated fake images, which cause major issues to the reliability of digital contents as a right representation of reality. Authenticity is the main problem in most digital image communications. Various forensic techniques have been developed for the verification of image integrity, authentication, and tampering detection. Digital image forensics aims at finding the authenticity of images by recovering history of the image. Image manipulations occur at the time of compression, which means changing the DCT coefficients. Forensic technique is capable of detecting chains of operators that are applied to an image. Here, the union of joint photographic expert’s group compression and full-frame linear filtering were studied and an accurate mathematical framework was derived for fully characterizing the probabilistic distributions of the discrete cosine transform (DCT) coefficients of the image, which get quantized and filtered. This statistical model is used for deriving a set of significant features from the DCT histogram of the input image; these features were fed to an effective classifier which effectively classifies different combinations of linear filtering and compression.

Keywords: Full-frame linear filtering, JPEG compression, linear classifier, image forensics

Cite this Article
Sreelakshmi M, Aswathy K. Recovering the History of Filtered JPEG Images for Forensic Detection. Journal of Software Engineering Tools & Technology Trends. 2016; 3(2): 1–6p.